bump sccache for linux x86_64 to allow caching while PGO'd

[klensy]

This bumps sccache for linux-only jobs to allow sccache work with PGO mozilla/sccache#952

This is improvement: 2h05m -> 1h44m (from 1 run) for dist-x86_64-linux #133033 (comment)

Why still use old release? Commit with improvement was old, so i used some release around. In future, more recent versions can be tested to see if there any improvements, while this one already gives one.

aarch64 update shouldn't change much, as there no PGO used.

For other OSes i will create separate PRs later.

r? infra

[klensy]

r? infra-ci

[klensy]

No promises about correct work with PGO :-)

[Kobzol]

r? @Kobzol

@bors try @rust-timer queue

[bors]

⌛ Trying commit 876d02e with merge a0fe848...

[mati865]

There is mozilla/sccache#1178 saying the first build will be fine, but later ones are failing. Definitely something to keep in mind if somebody notices spurious failures after merge.

I'm a bit curious how the versions were selected. Previously it was upgraded to the latest available version.
I've seen (in another project) some small but measurable performance improvement with some 0.7.x version due to the preprocessor caching.

[klensy]

Now I think that I can be wrong with time improvement, because compared try build with auto. Yes, unrolled build around 1h47m, sad.
Why old version? I mentioned that in first message, plus didn't want to mix few changes, like pgo, preprocessor, etc.

[klensy]

Well, in that case I will hold that PR and update to more recent version to check other features.

[mati865]

Oh shoot, I had read half of the description when I was interrupted. After coming back to it, I didn't remember to finish reading.

[bors]

☀️ Try build successful - checks-actions
Build commit: a0fe848 (a0fe8484f01f73841286024a032130e545185726)

[rust-timer]

Finished benchmarking commit (a0fe848): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (primary 2.9%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.9%	[2.2%, 3.6%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.9%	[2.2%, 3.6%]	2

Cycles

Results (primary 11.9%, secondary 2.6%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	11.9%	[2.5%, 25.2%]	13
Regressions ❌ (secondary)	2.6%	[1.9%, 3.3%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	11.9%	[2.5%, 25.2%]	13

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 788.599s -> 786.623s (-0.25%)
Artifact size: 335.45 MiB -> 335.42 MiB (-0.01%)

[Kobzol]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 876d02e with merge fc7d7eb...

[klensy]

some small but measurable performance improvement with some 0.7.x version due to the preprocessor caching.

Isn't direct mode for local only cache?

[mati865]

Right, in our setup, each builder keeps its own cache (local mode). This obviously works only with self hosted agents.

[bors]

☀️ Try build successful - checks-actions
Build commit: fc7d7eb (fc7d7eb691184e3f481bd02a1b0d8e790fe35cba)

[rust-timer]

Finished benchmarking commit (fc7d7eb): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (secondary 0.9%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.8%	[2.7%, 3.0%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.9%	[-2.9%, -2.9%]	1
All ❌✅ (primary)	-	-	0

Cycles

Results (primary 1.2%, secondary 2.6%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	1.2%	[0.9%, 1.6%]	6
Regressions ❌ (secondary)	2.6%	[2.4%, 2.8%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.2%	[0.9%, 1.6%]	6

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 788.11s -> 790.261s (0.27%)
Artifact size: 335.43 MiB -> 335.43 MiB (-0.00%)